Phylogenetic and morphological discord indicates introgressive hybridisation in two genera of Australian millipedes (Diplopoda, Polydesmida, Paradoxosomatidae)

Abstract Discord between molecular and morphological datasets was observed in two pairs of species of Australian millipedes in the family Paradoxosomatidae using morphological and molecular phylogenetic analysis (mitochondrial COI rDNA and 16 rRNA, and nuclear 28S rRNA). Close to the presumed distributional boundary between Pogonosternumnigrovirgatum (Carl, 1912) and Pogonosternumjeekeli Decker, 2017, near Dargo in Central Gippsland, Victoria, Pogonosternum specimens were collected which are phylogenetically closer to P.jeekeli in COI and 16S sequences, but are morphologically closer to P.nigrovirgatum. At Mount Osmond, Adelaide, South Australia, eight morphologically typical Somethuscastaneus (Attems, 1944) specimens were collected are phylogenetically closer to S.castaneus in 28S genealogy, but three of the eight are closer to S.lancearius Jeekel, 2002 in COI genealogy. These two cases are discussed in terms of hybridisation, past introgressive hybridisation events and aberrant morphology.

In Illinois, United States, the monotypic polydesmidan Illiniurus beattyi Shear, 1968 was later described as "transitional" in gonopod structure between Euryurus leachii Gray, 1832 and Auturus evides Bollman, 1887, both of which occur near the I. beattyi type locality (Jorgensen 2009). Later searches of the area for I. beattyi were unsuccessful, and Jorgensen (2009) was unable to determine whether the one known male of I. beattyi was a hybrid of E. leachii and A. evides, or simply an aberrant individual of one of these two euryurid species.
The introduction of genes from the gene pool of one species into that of another during hybridization, especially near species boundaries, is called introgressive hybridisation and can affect nuclear or mitochondrial DNA (Harper and Hart 2007, Harrison and Larson 2014, Toews and Brelsford 2012. Introgressive hybridisation has not yet been reported in the class Diplopoda.
Two cases of disagreement between relationships inferred from morphological similarity and molecular phylogenetics were observed in recent taxonomic studies of the Australian paradoxosomatid genera Pogonosternum Jeekel, 1965(Decker 2016a, Decker et al. 2017) and Somethus Chamberlin, 1920(Decker 2016b. These two cases are described here in detail and discussed with regard to hybridisation, past introgressive hybridisation events and aberrant morphology.

Specimen collection and preservation
Pogonosternum cf. nigrovirgatum (Carl, 1912)  Specimens were killed and stored in 95% ethanol, with a change of ethanol after 1-2 months. One male of Somethus castaneus from Mt. Osmond (SAM OM2138) was found dead in the field. DNA was not obtained from this specimen. The material is deposited in the Museums Victoria, Melbourne, Victoria, Australia (NMV), the South Australian Museum, Adelaide, Australia (SAM) and the Senckenberg Museum of Natural History Görlitz, Görlitz, Germany (SMNG).

Illustrations
Preserved specimens were imaged with a Leica Z6 Apo stereo microscope and Leica DFC420 camera. Focus-stacked images were assembled from 25-40 source images using the software package Leica Application Suite 4.5. All images were later edited using Adobe Photoshop CS4 and assembled into plates. The distribution maps were created with ArcMap 10.
Phylogenetic hypothesis was inferred for COI+16S, COI and 28S by using the maximum likelihood method conducted in MEGA6 (Tamura et al. 2011). The phylogenetic trees with the highest log likelihood (COI+16S: -5141; COI: -2565; 28S: -2328) are shown (Figs 3, 5). Initial trees for the heuristic search were obtained by applying the neighbour-joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach (Tamura et al. 2004). A discrete Gamma distribution was used to model evolutionary rate differences among sites (five categories (+G, parameter = COI+16S: 0.6793; COI: 0.1017; 28S: 0.0500)). The bootstrap consensus tree inferred from 1000 replicates (Felsenstein 1985) is here used as the best estimate of the phylogeny of each of the analysed taxa (Figs 3, 5). Mean uncorrected pairwise distances between terminals (transformed into percentages) were determined using MEGA6 (Tamura et al. 2011).

Molecular analysis
The final alignments consisted of 618 bp of COI mtDNA and 1206 bp of 28S rRNA in Somethus, and 1158 bp for COI+16S in Pogonosternum. Individual alignments are available upon request from the author. The best-fit model of nucleotide substitution selected using MrModelTest 2 was the General Time Reversible model with gamma distribution and proportion of invariant sites (Nei and Kumar 2000) for the individual COI and 16S dataset. The trees constructed from individual genes did not show significant conflicts in topology (nodes different among trees with support > 70% in ML) and no significant incongruence among the three genes was revealed by the ILD test (P > 0.81 in all of the pairwise comparisons), and the sequences were concatenated into a dataset comprised 1158 characters for phylogenetic analysis in Pogonosternum.

Morphology and sequence analysis of Pogonosternum cf. nigrovirgatum "Dargo"
Pogonosternum nigrovirgatum and P. jeekeli are very similar in somatic morphology, and the "Dargo" form agrees with both species in size, colouration, spiracle morphology and form of the leg pair 2 coxa in females. The "Dargo" form ( Fig. 1) agrees with typical P. nigrovirgatum (Fig. 2) in having an elongated gonopod femorite and differs from P. jeekeli in gonopods (Fig. 3) and male tarsal and tibial brushes only on leg pairs 1-7. See also Decker et al. (2017) for a detailed (re)description of P. nigrovirgatum and P. jeekeli and the considerable gonopod variability in these species as well as Decker 2016a for phylogeographic distribution of similar gonopod morphology. However, the four sequenced "Dargo" males are all phylogenetically closer to P. jeekeli in the combined COI and 16S dataset and form a well-supported group ( Fig. 3; bootstrap value 98% and with uncorrected percent difference of 2.4%), and appear in the tree next to a population of P. jeekeli from northeastern Tasmania. The Dargo collection site is close to the observed species distribution boundary between P. jeekeli and P. nigrovirgatum (Fig. 4).

Morphology and sequence analysis of Somethus castaneus from Mount Osmond
All six male specimens from Mt Osmond fully agree in gonopod morphology with S. castaneus and lack a medial prefemoral process (Decker 2016b). In the 28S phylogenetic analysis, all nine sequenced specimens (uncorrected p-distance of 0-0.08%) occur within a clade of S. castaneus and are closest to specimens from Onkaparinga, Horsnell Gully Conservation Park, and Brownhill Creek Recreation Park (Fig. 5). In the COI analysis (Fig. 5), three specimens from the northern slope of Mt Osmond (site S90) and three from the southwestern slope (site S110) occur with S. castaneus from Morialta Conservation Park, Brownhill Creek Recreation Park, Horsnell Gully Conservation Park, and Belair National Park (uncorrected p-distances 0.3-1.1%).
However, two specimens from site S90 and one from site S110 form a separate, wellsupported clade (100% bootstrap support) within Somethus, with genetic p-distances of 4.8-6.9% to S. lancearius, 5.3-6.9% to S. castaneus and 7.2-8.8 % to S. inflatus. The two Mt Osmond localities are in the centre of the S. castaneus distribution (Fig. 6).

Pogonosternum cf. nigrovirgatum "Dargo"
Despite the clear phylogenetic placement of this form within P. jeekeli as indicated by COI+16S gene trees (Fig. 3), the gonopod morphology of the seven Dargo specimens is closest to that of P. nigrovirgatum. In addition, tibial and tarsal brushes in the Dargo males extend only to leg pair 7, whereas in P. jeekeli the brushes reach leg pair 9 (74 studied males, Decker et al. 2017). In P. nigrovirgatum male brushes typically extend to leg pair 7, and infrequently to leg pair 9 in the area around Port Philip Bay. Pogonosternum nigrovirgatum and P. jeekeli form a sister clade (Decker 2016a) and the Dargo site is located close to the presumed boundary between the species (Fig. 4). The observed discordance between morphology and genetics could be the result of introgressive hybridisation of P. jeekeli and P. nigrovirgatum, resulting in a P. nigrovirgatum-type gonopod phenotype but a P. jeekeli mitochondrial COI and 16S genotypes. Furthermore, the finding of seven discordant males with a P. nigrovirgatumtype gonopod shows that the form is not the result of a developmental aberration in an individual within a P. jeekeli population. Another possibility is that P. cf. nigrovirgatum "Dargo" is simply a local variant of P. jeekeli, a species which is known to have variable gonopods in a genus that exhibits high gonopod variability (Decker 2016a, Decker et al. 2017). However, P. jeekeli is the only Pogonosternum species with tibial and tarsal brushes always ranging to leg pair 9. In contrast, brushes to leg pair 7 are found in P. adrianae Jeekel, 1982, P. montanum Decker, 2017, most males of P. nigrovirgatum and some males of P. laetificum Jeekel, 1982. This leg pair 7 limit is observed in four out of the five species in the genus Pogonosternum and likely to be plesiomorphic. Thus the hypothetical local variation in P. jeekeli resulting in the "Dargo" phenotype would involve both a change in gonopod morphology and a regression from apomorphy or common distribution of tibial and tarsal brushes in P. jeekeli. However, the author favours the hypothesis that two closely related Pogonosternum species have undergone introgressive hybridisation in an area of range overlap near Dargo. To distinguish between hypotheses of introgression versus gene tree lineage sorting, further gene sampling and analysis is needed of Pogonosternum individuals near the Dargo site and along near the presumed distribution boundary between P. nigrovirgatum and P. jeekeli. In addition, sampling of P. jeekeli and P. nigrovirgatum far from sympatry are needed.

Somethus castaneus from Mount Osmond
Somethus castaneus was sampled for sequencing at 13 localities covering most of the known species range, and genetic variability was shown to be low (up to 3.8% in uncorrected p-distances in COI) with three phylogenetic lineages (Decker 2016b). The morphologically typical S. castaneus from Mt Osmond, in the centre of the species range, includes three individuals whose COI sequences are not close to those of either the other S. castaneus, S. inflatus or S. lancearius, although they are slightly closer to those of S. lancearius (in % bp difference). Somethus lancearius is distributed in the north-eastern and eastern part of the Adelaide Hills, with some scattered, possibly introduced, occurrences of S. castaneus within its distributional area. The closest record of S. lancearius is about 17 km from Mt Osmond. Genetic variability within S. lancearius from five sampled localities is 1.6-5.8% (in uncorrected p-distances in COI) and with unique haplotypes that cor- respond to geographical areas (Decker 2016b). Widespread sampling of this species is no longer possible, as natural vegetation has largely been cleared within its range and S. lancearius is now restricted to scattered conservation areas and tiny remnants.
Several of the paratypes of S. inflatus were collected in the Adelaide suburb of Glen Osmond, near Mt Osmond, in 1969 (Jeekel 2002). These paratypes appear to have been lost and could not be compared with recently collected S. inflatus (Decker 2016b), and no S. inflatus have since been found on Mt Osmond. S. inflatus is distributed to the southwest of Adelaide on the Fleurieu Peninsula and on Kangaroo Island (Fig. 6). Jeekel's S. inflatus paratypes might represent a non-permanent introduction.
The three discordant S. castaneus found on Mt Osmond might be evidence for past introgression of mitochondrial DNA following hybridisation with another South Australian Somethus species. Alternatively, the anomalous individuals might represent a distinct and distantly related S. castaneus lineage which is either naturally occurring on Mt Osmond or introduced from another locality within the Adelaide Hills. There is no support for both hypotheses, but it seems that the likelihood of the presence (or former existence) on Mt Osmond of the in COI and 28S variable S. lancearius or S. inflatus is higher than that of a fourth distinct COI and 28S lineage in S. castaneus.
Future investigations with additional molecular markers and more individuals of from S. lancearius may not assist in clarifying the situation, as much of the former genetic variation of S. lancearius has probably been lost due to habitat loss and local extinctions. If the missing paratypes of S. inflatus are found in future, it might be possible to extract DNA and obtain sequences from them which could reveal whether S. inflatus in the Mt Osmond area has contributed mitochondrial COI to the local S. castaneus population.

Conclusion
The results presented here suggest that introgressive hybridisation may have occurred in the paradoxosomatid millipede genera Pogonosternum and Somethus in southeastern Australia. With the increasing use of molecular data in taxonomy and in barcoding projects, similar cases are likely to be found elsewhere. Interestingly, no evidence of introgressive hybridisation was found in more than 2000 COI sequences from Central European millipedes during the German Barcoding of Life Project (GBOL) (Wesener, Spelda, Reip, Decker pers. comm.). The phenomenon may be rare, or limited to narrow parapatric zones, as appears to be the case in Pogonosternum.